Solar Shape: An Indication of Future Solar Value

Solar photovoltaic (PV) capacity is set to dramatically increase in the coming years. EIA has forecast that U.S.-wide installed PV capacity will nearly triple to over 90 GW by 2030. Falling panel costs, government policy, and broadening acceptance of the need to de-carbonize power generation are driving this anticipated growth. But to achieve solar’s full potential, solar buyers and sellers need to understand how resource variability affects project value.

The economic value of a PV resource is determined by both its hourly power production and the hourly price of power. Historically, PV output has had a positive correlation with periods of above average power prices. However, PV resources have no marginal cost and, at high levels, can reduce power prices during periods of high solar output relative to load. As a result, increasing PV capacity may reduce the overall value of solar power. At the same time, power prices will become more sensitive to uncertain weather variables affecting solar output, including cloud cover and temperature. In short: the expected value of solar power will likely decrease, while the uncertainty in the value of solar power will increase.

Conventional approaches to resource assessment may miss one or both of these important effects. One conventional approach combines historical PV generation / price correlations with forecasted average power prices. Such an approach implicitly assumes that the co-dependence of solar output and power prices does not change over time. It misses both the decrease in expected value and the increase in uncertainty described above. A more sophisticated approach uses a power system optimization model, along with assumptions about increasing installations, to capture the decrease in expected value. But due to their complex data inputs and high computational costs, such power system optimization models are rarely iterated to capture uncertainty.

We take a third approach to modeling PV value. We begin with a streamlined power system fundamentals model that explicitly accounts for evolving capacity installations, fuel prices, and the demand for electricity over time. To account for the known limitations of such models in predicting hourly power prices, we train the output of our streamlined fundamentals model to improve out-of-sample hourly price prediction accuracy. We then extrapolate our trained fundamentals model to higher-solar futures under a variety of weather conditions. In this way we obtain both a point estimate, and an estimate of the uncertainty in the value of solar power.

Solar projects are long-lived assets that require significant upfront capital to be paid off over multiple decades. As a result, improved methods for estimating the value of solar power are critically important to the continued growth of the industry. Armed with such estimates, solar developers will be better able to obtain financing, manage their risks, and sell their power at the best possible price.